Google quantum computing on the cover of “Science” quantum simulation of chemical reactions


Today, Google’s quantum computer appeared on the cover of Science. They successfully simulated the isomerization reaction of diazene with 12 qubits.

This is the second time that Google Quantum Computer has appeared on the cover of top academic journals.

In October last year, Google’s quantum computer appeared on the cover of Nature because of its “quantum superiority”. It only took 200 seconds to solve the quantum circuit sampling problem that supercomputing took 10,000 years to solve.

What else can this quantum computer do? Google has said that it can simulate chemical molecules. They did it in less than a year.

Because molecules follow quantum mechanics, it is more reasonable to simulate with quantum calculations. You can calculate the properties of chemical substances with less calculation and information.

Quantum computers are very useful for simulating chemical molecules. In addition to Google, other companies with quantum computing technology are also studying, and Microsoft is one of them.

Last month, Microsoft published an article using quantum computing to help chemists find catalysts to convert carbon dioxide into formaldehyde. Demonstrates the application prospects of quantum computing and chemistry.

Quantum chemistry still has to use quantum computers

The Schrödinger equation is the basis of quantum chemistry and the basic law followed by chemical molecules. By solving the equation, the specific chemical properties of matter can be obtained.

But solving the Schrödinger equation is not easy. As the number of atoms in the molecule increases, the amount of computing to solve the equation increases exponentially.

Take the simpler benzene molecule (C6H6) in chemistry as an example. It has only 12 atoms, but its computational dimension reaches 1044, which cannot be processed by any supercomputer.

In order to simplify the solution process, long before the advent of computers, there were some approximation methods, such as the “Hartley-Fokker equation” used by Google. But even after simplification, the amount of calculation is huge. To make matters worse, in the chemical reaction process, that is, when the chemical bond is dissociated, the electronic structure of the molecular system will become more complicated, and it is difficult to perform relevant numerical calculations on any supercomputer.

In 2018, someone proposed a new quantum algorithm. The computational complexity is no longer exponential growth, but polynomial growth, which greatly reduces the computational difficulty.

If all the algorithms are available, a suitable quantum computer is missing.Google quantum computer simulates chemical reactions

Last year, Google’s Sycamore quantum processor achieved entanglement of 53 qubits, so I used it to simulate a few simple chemical molecules.

Google first calculates the binding energy of a hydrogen chain composed of 6 to 10 hydrogen atoms. The original method (yellow in the figure below) has a mediocre effect. After combining with algorithms such as VQE, the result obtained by the quantum computer is almost completely consistent with the real value.

The above is the static process of chemical molecules. Then, Google used Sycamore to simulate a simple chemical reaction: the isomerization of diazene.

The energy gap of diazene transition between cis and trans is 40.2 millihartris, and the result given by the quantum computer is 41±6 millihartris.

Although the accuracy is much worse than the previous simulation of the hydrogen atom chain, Google said that this is “the first time a quantum computer has been used to predict the mechanism of a chemical reaction.”

Ryan Babbush, the corresponding author of this article, said that although the above results can be simulated without a quantum computer, this work is still a big step forward for quantum computing.

In the future, this algorithm can be scaled up to simulate more complex reactions. To simulate the reaction of larger molecules, more qubits are needed.

Babbush believes that one day, we can even use quantum simulation to develop new chemical substances.

In order to simplify the solution process, long before the advent of computers, there were some approximation methods, such as the “Hartley-Fokker equation” used by Google. But even after simplification, the amount of calculation is huge.

To make matters worse, in the chemical reaction process, that is, when the chemical bond is dissociated, the electronic structure of the molecular system will become more complicated, and it is difficult to perform relevant numerical calculations on any supercomputer.

In 2018, someone proposed a new quantum algorithm. The computational complexity is no longer exponential growth, but polynomial growth, which greatly reduces the computational difficulty.

If all the algorithms are available, a suitable quantum computer is missing.

Reference link: